In Caenorhabditis elegans, the partitioning proteins (PARs), microfilaments (MFs), dynein, dynactin, and a nonmuscle myosin II all localize to the cortex of early embryonic cells. Both the PARs and the actomyosin cytoskeleton are required to polarize the anterior-posterior (a-p) body axis in one-cell zygotes, but it remains unknown how MFs influence embryonic polarity. Here we show that MFs are required for the cortical localization of PAR-2 and PAR-3. Furthermore, we show that PAR polarity regulates MF-dependent cortical forces applied to astral microtubules (MTs). These forces, which appear to be mediated by dynein and dynactin, produce changes in the shape and orientation of mitotic spindles. Unlike MFs, dynein, and dynactin, myosin II is not required for the production of these forces. Instead, myosin influences embryonic polarity by limiting PAR-3 to the anterior cortex. This in turn produces asymmetry in the forces applied to MTs at each pole and allows PAR-2 to accumulate in the posterior cortex of a one-cell zygote and maintain asymmetry.

First identified in the nematode Caenorhabditis elegans (Kemphues et al., 1988), the conserved partitioning proteins (PARs)* are required for cell polarity in many animal cell types (for review see Doe and Bowerman, 2001; Wodarz, 2002). In the one-cell stage C. elegans embryo, the PDZ domain protein PAR-3 and the Ring finger protein PAR-2 concentrate in complementary anterior and posterior cortical domains, respectively. Both are required to specify the anterior-posterior (a-p) body axis and to orient and position mitotic spindles relative to the a-p axis (Kemphues et al., 1988; Cheng et al., 1995; Etemad-Moghadam et al., 1995; Boyd et al., 1996) (Fig. 1, a and b).

One a-p asymmetry regulated by PAR-2 and PAR-3 appears during telophase of the first mitosis when the initially spherical posterior centrosome changes shape to form a disc, whereas the anterior centrosome remains spherical (Hill and Strome, 1988; Cheng et al., 1995; Keating and White, 1998). Previous observations of early embryos made using Nomarski DIC microscopy have suggested that PAR-3 inhibits flattening of the anterior spindle pole, whereas PAR-2 prevents this inhibition from occurring at the posterior pole (Cheng et al., 1995). Centrosome flattening may reflect an asymmetry in forces applied to the centrosomes through astral microtubules (MTs) that contact the cell cortex during mitosis. This asymmetry in force displaces the first mitotic spindle toward the posterior pole and generates lateral rocking motions of the posterior spindle pole during the asymmetric first division of a one-cell zygote (Cheng et al., 1995; Grill et al., 2001; Tsou et al., 2002).

SPD-5 and centrosome shape

To examine centrosome shape, we stained fixed wild-type and par mutant embryos with antibodies that recognize a centrosomal protein called SPD-5 (Hamill et al., 2002) (Fig. 1, c, f, and i). In wild-type, the posterior centrosome flattened late in mitosis, whereas the anterior centrosome remained spherical (Fig. 1 c). In par-2(lw32) mutant embryos, PAR-3 spread around the posterior cortex, and both centrosomes remained spherical, like the anterior centrosome in a wild-type zygote (Fig. 1, d–f) (Cheng et al., 1995; Etemad-Moghadam et al., 1995). Conversely, PAR-2 accumulated throughout the cortex in par-3(it71) mutants, and both centrosomes flattened to resemble a wild-type posterior centrosome (Fig. 1, g–i) (Cheng et al., 1995; Boyd et al., 1996).

Centrosome flattening requires microfilaments

Disruption of MF assembly results in a-p polarity defects similar to those caused by mutations in par-2. In wild-type embryos treated with cytochalasin D (Hill and Strome, 1988) or latrunculin A (LatA; Fig. 2 c; eight out of nine embryos), neither centrosome flattened. The failure of either pole to flatten could result from mislocalized PAR-3 inhibiting flattening at both poles as in par-2 mutants (Cheng et al., 1995). Moreover, MFs might be required for cortical localization of the PAR proteins, with such localization being important for their function. Therefore, we examined the localization of PAR-2 and PAR-3 in embryos exposed to LatA. We found that PAR-2 and PAR-3 both require intact MFs to localize to the cortex. Both were undetectable at the cortex, or present at severely reduced levels, in the presence of LatA (Fig. 2, a and b; n ≥ 5 for each; see Materials and methods). PAR-2 accumulated around the centrosomes of LatA-treated embryos as was observed recently in pod mutants with defects in a-p polarity (Rappleye et al., 2002). We also examined centrosome flattening and PAR localization in embryos with reduced levels of the profilin PFN-1, which we have recently shown is required for the assembly of cortical MFs (Severson et al., 2002) (Fig. 2 g). Consistent with our findings in LatA-treated embryos, the posterior centrosome failed to flatten in embryos depleted of PFN-1 using dsRNA-mediated gene silencing, or RNAi (Fig. 2 g), and PAR-2 was undetectable at the cortex but instead localized around centrosomes (Fig. 2 f; 10 out of 12 embryos). Although PAR-3 was always detected at the cortex in PFN-1–depleted embryos, it was present at much reduced levels compared with wild-type embryos fixed on the same slides (Fig. 2 e; six out of six embryos). The remaining cortical PAR-3 may simply reflect residual MF assembly because low levels of cortical F-actin still assemble in embryos with reduced levels of profilin (Severson et al., 2002). We conclude that centrosome flattening and the cortical localization of PAR-2 and PAR-3 all require an intact MF cytoskeleton.

PAR-3 prevents flattening of the anterior centrosome in wild-type embryos and of both the anterior and posterior centrosomes in par-2 mutants. In LatA-treated embryos, PAR-3 is not present at the cortex, but cytoplasmic PAR-3 could still function to prevent centrosome flattening. Therefore, we examined centrosome shapes in par-3 mutants exposed to LatA. We found that both centrosomes, which are flattened in par-3 single mutants and in par-2 par-3 double mutants, were spherical as in LatA-treated wild-type embryos (Fig. 2 d; n = 6). Both centrosomes were also spherical in pfn-1; par-3 double mutant embryos (Fig. 2 h; n = 10). We conclude that in addition to being required for the cortical localization of PAR-2 and PAR-3, MFs are required for the process of centrosome flattening itself. PAR-2 and PAR-3 together restrict flattening to the posterior pole but are not required for production of the forces that underlie this process.

We also tested whether MFs are required for spindle orientation in two-cell stage embryos. In wild-type, the centrosomes of both two-cell stage blastomeres initially are aligned orthogonal to the a-p axis. During mitotic prophase, the nucleus and its associated centrosomes (nucleocentrosomal complex [NCC]) in the posterior blastomere rotate 90°. A mitotic spindle subsequently assembles parallel to the a-p axis in this cell, whereas the spindle in the anterior cell remains transverse (Hyman and White, 1987) (Fig. 2 i). In par-3 mutants, both NCCs rotate (Fig. 2 k; eight out of ten embryos), whereas both remain transverse in par-2 mutants (Kemphues et al., 1988). However, both rotate in par-2 par-3 double mutant embryos, indicating that neither PAR protein is required for NCC rotation (Cheng et al., 1995). As shown previously in experiments using cytochalasin D (Hyman and White, 1987), we observed that the posterior NCC failed to rotate in wild-type two-cell stage embryos treated with LatA (Fig. 2 j; n = 6; see Materials and methods). Similarly, both NCCs failed to rotate in two-cell stage par-3 mutant embryos exposed to LatA (Fig. 2 l, n = 6). Thus, MFs mediate changes both in spindle pole shape at the one-cell stage and in spindle orientation at the two-cell stage, with PAR-2 and PAR-3 regulating where these changes occur.

Myosin II is not required for centrosome flattening

We next examined how myosin II influences the MF-dependent forces that flatten spindle poles. Depletion of the nonmuscle myosin II heavy chain (NMY)-2 or of the myosin II regulatory light chain (MLC)-4 (Guo and Kemphues, 1996; Shelton et al., 1999) results in embryonic polarity defects similar to those in LatA-treated embryos: the first mitotic spindle remains centrally positioned and both spindle poles remain spherical (Fig. 3, b and f). However, one difference is that PAR-3 accumulates around both the anterior and posterior cortex in embryos depleted of either myosin II subunit (Guo and Kemphues, 1996; Shelton et al., 1999), whereas PAR-3 is not present at the cortex in LatA-treated embryos (see above). In contrast to PAR-3, PAR-2 was usually present in a reduced cortical patch in mutant embryos depleted of NMY-2 or MLC-4, (Fig. 3, a and e; five out of eight nmy-2 and six out of eight mlc-4 embryos) or was undetectable at the cortex (three out of eight nmy-2 and two out of eight mlc-4 mutants) (Shelton et al., 1999). Thus, unlike MFs, neither NMY-2 nor MLC-4 are required for PAR-3 to associate with the cortex, but they are required for the polarized distribution of cortical PAR-3 and for the posterior cortical localization of PAR-2.

We next asked whether the PAR-3 present throughout the cortex in myosin-depleted embryos inhibits flattening at both spindle poles. We found that both centrosomes flattened in nmy-2(RNAi); par-3(it71) (n = 24) and mlc-4(RNAi)s; par-3(it71) double mutant embryos (Fig. 3, d and h; n = 7). Thus, NMY-2 and MLC-4 are dispensable for the MF-dependent forces that underlie centrosome flattening. Instead, myosin II appears to facilitate the establishment of normal, complementary domains of PAR-2 and PAR-3, which in turn regulate the distribution of the forces that influence spindle shape and position.

Myosin II restricts PAR-3 to the anterior cortex

As described above, PAR-3 accumulates around the cortex of myosin-depleted embryos, whereas PAR-2 and PAR-3 localize in mutually exclusive cortical domains in wild-type zygotes. Myosin II could influence the localization of PAR-2 and PAR-3 by facilitating expansion of the PAR-2 domain, thereby restricting PAR-3 to the anterior cortex. Alternatively, myosin might limit PAR-3 localization to the anterior hemisphere, thus permitting expansion of the PAR-2 domain. To distinguish between these two models, we examined the localization of PAR-2 in NMY-2–depleted and in MLC-4–depleted par-3 mutant embryos. In both cases, we found that PAR-2 was present throughout the cortex, suggesting that neither myosin II subunit is required for cortical localization or expansion of PAR-2 (Fig. 3, c and g; n ≥ 5 for each double mutant). Instead, myosin appears to restrict PAR-3 to the anterior, with ectopic PAR-3 preventing PAR-2 accumulation at the cortex in myosin II–depleted embryos.

Centrosome flattening requires dynein and dynactin

The results described above suggest that MFs either recruit or activate a cortical motor protein that pulls on astral MTs to influence the shape and position of mitotic spindles. Both the dynactin complex and the minus end–directed MT motor dynein localize to the cortex of early embryos, and spindle rotation fails in two-cell stage embryos in which the dynein–dynactin complex has been partially depleted by RNA interference (Skop and White, 1998; Gönczy et al., 1999). Further reducing dynein–dynactin function disrupts pronuclear migration and the assembly and orientation of the first mitotic spindle (Gönczy et al., 1999). To determine whether dynein and dynactin are required for centrosome flattening in one-cell stage embryos, we partially depleted either the dynein heavy chain DHC-1 or a C. elegans orthologue of the dynactin component p150glued DNC-1 (see Materials and methods). The posterior centrosome remained spherical in all DNC-1–depleted embryos in which the first mitotic spindle rotated to lie along the a-p axis (Fig. 4 b; n = 9). Similarly, we observed spherical centrosomes in some DHC-1–depleted embryos (Fig. 4 c; 4 out of 20 embryos; 4 embryos exhibited defects in chromosome segregation and in centrosome flattening, whereas 16 embryos appeared wild type during the first mitotic division). Exposure of embryos to low doses of nocodazole that shorten but do not eliminate MTs also disrupted centrosome flattening (Fig. 4 d; five out of seven embryos). We conclude that both dynein function and contact between astral MTs and the cortex are required for centrosome flattening.

Concluding remarks

Our data suggest that the nonmuscle myosin II subunits NMY-2 and MLC-4 mediate only a subset of F-actin–dependent processes during polarization of the a-p axis in a C. elegans zygote. F-actin is required for at least four polarity functions in the one-cell stage embryo: the cortical localizations of PAR-2, PAR-3, and NMY-2, and centrosome flattening (Fig. 5 a). In contrast, NMY-2 and MLC-4 are dispensable for cortical PAR localization and for centrosome flattening. Myosin II instead restricts PAR-3 to the anterior cortex, which permits expansion of the PAR-2 domain. As ectopic PAR-3 accumulates in the posterior of par-2 single mutants, myosin is not sufficient to restrict PAR-3. Thus, both PAR-2 and myosin II are required to limit PAR-3 to the anterior cortex.

The flattening of the posterior centrosome along the transverse axis may occur as a result of cortical forces that are applied to astral MTs and displace the first mitotic spindle toward the posterior pole (Grill et al., 2001; Tsou et al., 2002). The net magnitude of these forces is greater in the posterior hemisphere, and the posterior pole of the first mitotic spindle rocks from side to side during telophase. Thus, lateral forces act on astral MTs that contact the posterior cortex late in mitosis when centrosome flattening is observed. Because both spindle poles exhibit rocking motions in par-3 mutant embryos, whereas neither pole shows rocking in par-2 mutants, normal PAR polarity appears necessary to restrict lateral forces to the posterior pole (Cheng et al., 1995). Our data suggest that MFs recruit the dynein–dynactin complex to the cortex to apply these lateral forces to astral MTs. NMY-2 and MLC-4 are required for a polarized distribution of the PAR proteins, which in turn regulate the localization or the function of the dynein–dynactin motor complex, thus influencing both the position and shape of the first mitotic spindle (Fig. 5 b).

Recently, two models have been proposed to explain the establishment of asymmetry in the forces that position mitotic spindles in C. elegans zygotes. First, a DEP domain protein called LET-99 accumulates in a cortical stripe that is displaced toward the posterior pole, and high levels of LET-99 have been proposed to attenuate dynein-dependent forces applied to astral MTs that contact the cell cortex (Tsou et al., 2002). Properly positioned lateral attenuation would lower forces that normally oppose those applied to the spindle pole from the posterior-most cortex, producing a greater net force toward the posterior (see Fig. 7 in Tsou et al., 2002).

Alternatively, it has been suggested that MFs are unlikely to be involved in generating the cortical forces that act on spindle poles (Hill and Strome, 1988; Grill et al., 2001). This conclusion is based on experiments in which brief pulses of cytochalasin D, applied and washed out before anaphase, were sufficient to prevent posterior displacement of the first mitotic spindle during anaphase. Furthermore, cytochalasin D pulses applied during anaphase did not prevent posterior displacement (Hill and Strome, 1988). These findings suggest that MFs are not directly required for posterior displacement of the first mitotic spindle. Grill et al. (2001) therefore suggested that increased astral MT instability associated with the posterior cortex might account for the greater net posterior force. For example, such instability might facilitate pulling of the spindle pole toward the posterior cortex as astral MTs shorten.

Our findings support a role for MFs and the dynein–dynactin motor complex in applying forces to spindle poles via astral MTs that contact the cell cortex. It is possible that the pulses of cytochalasin D used by Hill and Strome (1988) were sufficient to disrupt some aspects of polarity but not to disrupt dynein–dynactin-mediated application of forces to astral MTs. Alternatively, cytochalasin D pulses may fully disrupt MF function, but two different force mechanisms could operate during spindle positioning. MT instability might account for posterior displacement, with dynein–dynactin forces generating only lateral rocking and flattening of the posterior spindle pole. In support of this possibility, we sometimes observed an absence of spindle pole flattening even though the spindle was displaced normally toward the posterior pole (Fig. 4). MF function is not limited to spindle flattening and rocking though, because MFs, dynein, and dynactin also are required for spindle rotation at the two-cell stage in wild-type and par-3 mutant embryos. Finally, MT asters undergo abnormal lateral rocking movements early in mitosis in one-cell let-99 mutant embryos, and this abnormal rocking also requires dhc-1 (Tsou et al., 2002). We conclude that dynein–dynactin-mediated forces exert an extensive influence on mitotic spindle positioning in early C. elegans embryos.

Strains and alleles

C. elegans strains were cultured as described previously; N2 Bristol was used as the wild-type strain (Brenner, 1974). The following alleles and balancer chromosomes were used in this study: LGIII: par-2(lw32), unc-45(e286ts), lon-1(e185), par-3(it71), sC1[dpy-1(e1) let], and qC1[dpy-19 glp-1].

Immunofluorescence and microscopy

Embryos were fixed and stained as described (Severson et al., 2000). Antibodies were diluted in PBS containing 3% BSA as follows: PAR-2, 1:20; PAR-3, 1:10; SPD-5, 1:1,000; and antiactin (ICN), 1:100. DNA was labeled with a 10-min incubation in 0.2 μM TOTO-3 (Molecular Probes). Images were acquired using a Radiance laser-scanning confocal microscope (Bio-Rad Laboratories). For observations of centrosome shape following dhc-1(RNAi), dnc-1(RNAi), and nocodazole treatment, embryos expressing a histone–GFP and a β-tubulin–GFP fusion (Praitis et al., 2001) were mounted on a 4% agarose cushion and observed using a spinning disc confocal microscope (PerkinElmer).

LatA and nocodazole exposure

For LatA treatment, embryos were permeabilized by laser ablation of the eggshell (Severson et al., 2002) or by gentle pressure (Hill and Strome, 1988). Embryos were permeabilized during pronuclear migration for observations of centrosome flattening or after the completion of cytokinesis I for observations of spindle orientation in two-cell embryos. Embryos were incubated for at least 10 min in culture medium containing 100 μM LatA (prepared from a 10 mM stock in DMSO) or in DMSO as a control and then observed by Nomarski microscopy or fixed and processed for immunocytochemistry. Centrosome flattening, spindle orientation, and PAR localization were normal in DMSO-treated embryos, and cell cycle progression continued normally in both LatA- and DMSO-treated embryos. For nocodazole treatment, prepronuclear migration stage embryos were mounted on a 4% agarose cushion under a coverslip and then bathed in 20 μg/ml nocodazole in M9 prepared from a 1 mg/ml stock solution in DMSO.


Double stranded RNA was prepared and injected by standard methods (Fire et al., 1998). The following cDNA clones were used as templates: mlc-4, yk167f10; nmy-2, yk45d7; and pfn-1, yk402e3. Sequences corresponding to dhc-1 and dnc-1 were amplified from genomic DNA using the following primers: dhc-1: aaggaaggagctcaacgaca, cctttccttcctgggtcttc; and dnc-1: tcatcgaatccttccgtttc, gaagcacgcggttgatttat. PCR products were cloned into PCRII-TOPO (Invitrogen), and single stranded RNA was transcribed with T7 polymerase, injected at a concentration of 1 mg/ml, and embryos were analyzed 18–20 h postinjection.

We would like to thank Ken Kemphues and Michael Glotzer for generously providing antibodies used in this study, Yuji Kohara for cDNA clones, Phil Crews for LatA, Barbara Meyer for patience and for her facilities while doing final experiments for this manuscript, the reviewers for suggested experiments and text changes, and the C. elegans Genetics Center (funded by the National Institutes of Health National Center for Research Resources) for strains.

A.F. Severson and B. Bowerman were funded by the National Institutes of Health (R01GM049869).


Abbreviations used in this paper: a-p, anterior-posterior; DHC, dynein heavy chain; DNC, dynactin component p150glued; LatA: latrunculin A; MF, microfilament; MLC, myosin II regulatory light chain; MT, microtubule; NCC, nucleocentrosomal complex; NMY, nonmuscle myosin II heavy chain; PAR, partitioning protein; RNAi, double stranded RNA-mediated interference.

Boyd, L., S. Guo, D. Levitan, D.T. Stinchcomb, and K.J. Kemphues.
. PAR-2 is asymmetrically distributed and promotes association of P granules and PAR-1 with the cortex in C. elegans embryos.
Brenner, S.
. The genetics of Caenorhabditis elegans.
Doe, C.Q., and B. Bowerman.
. Asymmetric cell division: fly neuroblast meets worm zygote.
Curr. Opin. Cell Biol.
Cheng, N.N., C.M. Kirby, and K.J. Kemphues.
. Control of cleavage spindle orientation in Caenorhabditis elegans: the role of the genes par-2 and par-3.
Etemad-Moghadam, B., S. Guo, and K.J. Kemphues.
. Asymmetrically distributed PAR-3 protein contributes to cell polarity and spindle alignment in early C. elegans embryos.
Fire, A., S. Xu, M.K. Montgomery, S.A. Kostas, S.E. Driver, and C.C. Mello.
. Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans.
Gönczy, P., S. Pichler, M. Kirkham, and A.A. Hyman.
. Cytoplasmic dynein is required for distinct aspects of MTOC positioning, including centrosome separation, in the one cell stage Caenorhabditis elegans embryo.
J. Cell Biol.
Grill, S.W., P. Gönczy, E.H. Stelzer, and A.A. Hyman.
. Polarity controls forces governing asymmetric spindle positioning in the Caenorhabditis elegans embryo.
Guo, S., and K.J. Kemphues.
. A non-muscle myosin required for embryonic polarity in Caenorhabditis elegans.
Hamill, D.R., A.F. Severson, J.C. Carter, and B. Bowerman.
. Centrosome maturation and mitotic spindle assembly in C. elegans require SPD-5, a protein with multiple coiled-coil domains.
Dev. Cell.
Hill, D.P., and S. Strome.
. An analysis of the role of microfilaments in the establishment and maintenance of asymmetry in Caenorhabditis elegans zygotes.
Dev. Biol.
Hyman, A.A., and J.G. White.
. Determination of cell division axes in the early embryogenesis of Caenorhabditis elegans.
J. Cell Biol.
Keating, H.H., and J.G. White.
. Centrosome dynamics in early embryos of Caenorhabditis elegans.
J. Cell Sci.
Kemphues, K.J., J.R. Priess, D.G. Morton, and N.S. Cheng.
. Identification of genes required for cytoplasmic localization in early C. elegans embryos.
Praitis, V., E. Casey, D. Collar, and J. Austin.
. Creation of low-copy integrated transgenic lines in Caenorhabditis elegans.
Rappleye, C.A., A. Tagawa, R. Lyczak, B. Bowerman, and R.V. Aroian.
. The anaphase-promoting complex and separin are required for embryonic anterior-posterior axis formation.
Dev. Cell.
Severson, A.F., D.R. Hamill, J.C. Carter, J. Schumacher, and B. Bowerman.
. The aurora-related kinase AIR-2 recruits ZEN-4/CeMKLP1 to the mitotic spindle at metaphase and is required for cytokinesis.
Curr. Biol.
Severson, A.F., D.L. Baillie, and B. Bowerman.
. A formin homology protein and a profilin are required for cytokinesis and Arp2/3-independent assembly of cortical microfilaments in the early Caenorhabditis elegans embryo.
Curr. Biol.
Shelton, C.A., J.C. Carter, G.C. Ellis, and B. Bowerman.
. The nonmuscle myosin regulatory light chain gene mlc-4 is required for cytokinesis, anterior-posterior polarity, and body morphology during Caenorhabditis elegans embryogenesis.
J. Cell Biol.
Skop, A.R., and J.G. White.
. The dynactin complex is required for cleavage plane specification in early Caenorhabditis elegans embryos.
Curr. Biol.
Tsou, M.F., A. Hayashi, L.R. DeBella, G. McGrath, and L.S. Rose.
. LET-99 determines spindle position and is asymmetrically enriched in response to PAR polarity cues in C. elegans embryos.
Wodarz, A.
. Establishing cell polarity in development.
Nat. Cell Biol.